We have previously described a method for dealing with missing data in a prospective cardiac registry initiative. The method involves merging registry data to corresponding ICD-9-CM administrative data to fill in missing data 'holes'. Here, we describe the process of translating our data merging solution to ICD-10, and then validating its performance.
A multi-step translation process was undertaken to produce an ICD-10 algorithm, and merging was then implemented to produce complete datasets for 1995-2001 based on the ICD-9-CM coding algorithm, and for 2002-2005 based on the ICD-10 algorithm. We used cardiac registry data for patients undergoing cardiac catheterization in fiscal years 1995-2005. The corresponding administrative data records were coded in ICD-9-CM for 1995-2001 and in ICD-10 for 2002-2005. The resulting datasets were then evaluated for their ability to predict death at one year.
The prevalence of the individual clinical risk factors increased gradually across years. There was, however, no evidence of either an abrupt drop or rise in prevalence of any of the risk factors. The performance of the new data merging model was comparable to that of our previously reported methodology: c-statistic = 0.788 (95% CI 0.775, 0.802) for the ICD-10 model versus c-statistic = 0.784 (95% CI 0.780, 0.790) for the ICD-9-CM model. The two models also exhibited similar goodness-of-fit.
The ICD-10 implementation of our data merging method performs as well as the previously-validated ICD-9-CM method. Such methodological research is an essential prerequisite for research with administrative data now that most health systems are transitioning to ICD-10.
The structure of the Canadian health care system lends itself to health services and health outcomes research. It is possible to track hospital admissions and discharges, physician billings and prescriptions using administrative databases. In addition, several provinces have developed registries that provide detailed clinical and procedural information. Using the unique personal health numbers assigned to all Canadian residents, linkage between administrative databases and population-based clinical registries provides important information regarding the use of health services and health outcomes.
To determine the extent of cross-border (British Columbia-Alberta border) use of cardiac services by British Columbia residents.
Population rates of cardiac procedures were calculated using two prospective clinical registries (British Columbia Cardiac Registries and Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease [APPROACH]), as well as administrative databases (the British Columbia Ministry of Health's hospitalization and Medical Services Plan databases).
Analyses using only British Columbia data suggest low cardiac procedure rates for patients living in eastern British Columbia. By accessing APPROACH data, it was determined that more than 80% of British Columbia cardiac patients living along the British Columbia-Alberta border access procedural services in Alberta.
While residents of eastern British Columbia appear to have reduced access to cardiac services when data from British Columbia are analyzed in isolation, they are actually accessing care in Alberta. Analyses based solely on single province data sources will underestimate cardiac procedures rates.
Surnames have the potential to accurately identify ancestral origins as they are passed on from generation to generation. In this study, we developed and validated a Chinese surname list to define Chinese ethnicity.
We conducted a literature review, a panel review, and a telephone survey in a randomly selected sample from a Canadian city in 2003 to develop a Chinese surname list. The list was then validated to data from the Canadian Community Health Survey. Both surveys collected information on self-reported ethnicity and surname.
Of the 112,452 people analyzed in the Canadian Community Health Survey, 1.6% were self-reported as Chinese. This was similar to the 1.5% identified by the surname list. Compared with self-reported Chinese ethnicity (reference standard), the surname list had 77.7% sensitivity, 80.5% positive predictive value, 99.7% specificity, and 99.6% negative predictive value. When stratifying by sex and marital status, the positive predictive value was 78.9% for married women and 83.6% for never married women.
The Chinese surname list appears to be valid in identifying Chinese ethnicity. The validity may depend on the geographic origins and Chinese dialects in given populations.
Despite evidence of regional variation across North America, there have been no comprehensive studies of cardiac procedure rates for coronary heart disease in Canada.
To use available administrative data and a survey of catheterization facilities to examine regional and demographic variations in cardiovascular procedure rates.
A survey of all cardiac catheterization facilities in Canada was conducted, and the procedure counts from these facilities were used to determine provincial catheterization rates from 1997/1998 to 2001/2002. Procedure counts for 1997/1998 to 1999/2000 for coronary artery bypass graft (CABG) surgery and percutaneous coronary intervention (PCI) were provided by the Canadian Institute for Health Information and used to calculate revascularization procedure rates. Population projections provided by Statistics Canada were used as denominators for calculating the rates, and direct standardization was used to obtain age- and sex-adjusted rates.
The crude rate of cardiac catheterization in Canada increased from 359.9 to 471.5 per 100,000 population across the five years studied. There was considerable variation in revascularization procedure rates across health regions and provinces. Between 1997/1998 and 1999/2000, there was little increase in the rate of CABGs performed in Canada but a marked increase in the rate of PCIs. For both CABG and PCI, rates were higher for men than women, and highest in the 65- to 74-year-old age category. This study provides a valuable 'snapshot' of cardiac procedure use rates but indicates a clear need for more comprehensive collection of cardiac care data in Canada.
We validated the accuracy of case definitions for hypertension derived from administrative data across time periods (year 2001 versus 2004) and geographic regions using physician charts. Physician charts were randomly selected in rural and urban areas from Alberta and British Columbia, Canada, during years 2001 and 2004. Physician charts were linked with administrative data through unique personal health number. We reviewed charts of approximately 50 randomly selected patients >35 years of age from each clinic within 48 urban and 16 rural family physician clinics to identify physician diagnoses of hypertension during the years 2001 and 2004. The validity indices were estimated for diagnosed hypertension using 3 years of administrative data for the 8 case-definition combinations. Of the 3,362 patient charts reviewed, the prevalence of hypertension ranged from 18.8% to 33.3%, depending on the year and region studied. The administrative data hypertension definition of "2 claims within 2 years or 1 hospitalization" had the highest validity relative to the other definitions evaluated (sensitivity 75%, specificity 94%, positive predictive value 81%, negative predictive value 92%, and kappa 0.71). After adjustment for age, sex, and comorbid conditions, the sensitivities between regions, years, and provinces were not significantly different, but the positive predictive value varied slightly across geographic regions. These results provide evidence that administrative data can be used as a relatively valid source of data to define cases of hypertension for surveillance and research purposes.
This study assessed the validity of a widely-accepted administrative data surveillance methodology for identifying individuals with diabetes relative to three laboratory data reference standard definitions for diabetes.
We used a combination of linked regional data (hospital discharge abstracts and physician data) and laboratory data to test the validity of administrative data surveillance definitions for diabetes relative to a laboratory data reference standard. The administrative discharge data methodology includes two definitions for diabetes: a strict administrative data definition of one hospitalization code or two physician claims indicating diabetes; and a more liberal definition of one hospitalization code or a single physician claim. The laboratory data, meanwhile, produced three reference standard definitions based on glucose levels +/- HbA1c levels.
Sensitivities ranged from 68.4% to 86.9% for the administrative data definitions tested relative to the three laboratory data reference standards. Sensitivities were higher for the more liberal administrative data definition. Positive predictive values (PPV), meanwhile, ranged from 53.0% to 88.3%, with the liberal administrative data definition producing lower PPVs.
These findings demonstrate the trade-offs of sensitivity and PPV for selecting diabetes surveillance definitions. Centralized laboratory data may be of value to future surveillance initiatives that use combined data sources to optimize case detection.